THE MORAL STATUS OF ENABLING HARM Samuel C.

Rickless [To appear in Pacific Philosophical Quarterly] Abstract According to the Doctrine of Doing and Allowing, it is more difficult to justify doing harm than it is to justify allowing harm. Enabling harm consists in withdrawing an obstacle that would, if left in place, prevent a pre-existing causal sequence from leading to foreseen harm. There has been a lively debate concerning the moral status of enabling harm. According to some (e.g., McMahan, Vihvelin and Tomkow), many cases of enabling harm are morally indistinguishable from doing harm. Others (e.g., Foot, Hanser) support the Equivalence Hypothesis, according to which enabling harm is morally equivalent to allowing harm. Here I argue that there every reason to embrace, and no reason to reject, the Equivalence Hypothesis.

1. INTRODUCTION There is a puzzle at the heart of non-consequentialism that has not yet received a completely satisfying solution. The puzzle concerns the moral status of what Philippa Foot has described as enabling harm, i.e., withdrawing an obstacle that would, if left in place, prevent a pre-existing causal sequence from leading to foreseen harm. Nonconsequentialists who take the distinction between doing and allowing harm to be morally relevant are divided about the proper moral classification of enabling harm. For some, all cases of enabling harm are morally equivalent to cases of allowing harm.1 For others, at least some cases of enabling harm are morally equivalent to cases of doing harm.2 The latter claim has also been endorsed by consequentialists, who treat it as a

point in favor of their approach to ethics that there appears to be no principled nonconsequentialist way of explicating the moral relevance of enabling cases.3 The purpose of this paper is to clear up the existing confused state of non-consequentialist theorizing on this topic, and settle the debate by vindicating Foots original and far-seeing proposal.4

2. THE DOCTRINE OF DOING AND ALLOWING In a set of closely related publications, Philippa Foot describes what one might call the doctrine of doing and allowing (or DDA, for short).5 According to the DDA, the distinction roughly captured in ordinary discourse by the terms doing harm and allowing harm is morally relevant in that the duty not to do harm is stronger than the duty not to allow it to occur. Paradigm cases of doing harm, on Foots view, consist in initiating or sustaining a causal sequence that leads to (foreseen) harm. To initiate a causal sequence is to set it in motion; to sustain it is to keep it going when it would otherwise have stopped. Thus, for example, if I set a trolley in motion in such a way that the trolley crushes a person who is trapped on the trolley tracks, then I have done harm by initiating a causal sequence that leads to it. And if I depress the trolleys accelerator pedal as the trolley is slowing down in such a way that the additional impetus results in the crushing of the trapped person, then I have done harm by sustaining a causal sequence that leads to it. Paradigm cases of allowing harm, on the other hand, consist in failing to prevent a pre-existing causal sequence from bringing about (foreseen) harm. (Henceforth, for the sake of convenience, I will drop the reference to the harms being foreseen.) Thus, for example, if I could divert a speeding trolley that is bearing down on

one trapped person onto an empty side-track but fail to do so, then I allow harm to befall the trapped person by failing to prevent the trolley from crushing him.6 To say, as Foot does, that the duty not to do harm is stronger than the duty not to allow it to occur is to say, among other things, that when all else is equal, in situations in which one is forced to choose between doing harm and allowing harm (to persons who do not wish to suffer the harm)7, one is morally required to allow harm unless the harm that is allowed is out of all proportion to the harm that would otherwise be done. In particular, if one is forced to choose between doing harm to one and allowing the same kind of harm to befall five, one is morally required to avoid doing harm. A case of this kind is Rescue:

Rescue Roger is driving across a mountainous region to save five innocent people who are being slowly crushed to death by a large pneumatic press. If Roger reaches the press, he will be able to shut it off and the five will walk away unharmed. As Roger reaches a narrow mountain pass, he discovers that there is one innocent person trapped on the road ahead.8

In Rescue, Roger is forced to choose between two options. The first option is to stop the car, in which case the one lives and the five die. The second option is to drive across the mountain pass just in time to shut off the pneumatic press, in which case the one dies and the five live. It seems clear that it is morally impermissible for Roger to choose the second option over the first: he may not crush the one on his way to saving the five from

being crushed. This kind of forced choice case brings out in the clearest possible way the inconsistency between the DDA and consequentialism, a theory that bids us to produce, from among all available alternatives, the one that leads to the best consequences overall. At the same time, if the only alternative to Rogers crushing the one is the failure to save a sufficiently large or disproportionate number (say, one million), then it may, on Foots account, be permissible for Roger to crush the one. The duty not to do harm is therefore more stringent than the duty not to allow harm to occur, but it is not absolute.

3. ENABLING HARM Within a non-consequentialist theory that includes the DDA, what is the moral status of enabling harm? One view, championed by Foot herself, is that harmful enablings are morally on a par with harmful allowings. Foot offers us the case of respirator removal as evidence for the claim that the removal of an obstacle that stands in the way of a preexisting potentially harmful causal sequence is morally equivalent to the failure to prevent some pre-existing causal sequence from leading to harm. (Call this claim the Equivalence Hypothesis.) And, at least at first blush, this seems exactly right. One way to test the Equivalence Hypothesis is by considering cases in which one has no better choice than between allowing harm and enabling harm, where the only additional putatively morally relevant difference between the two options is that the consequences of allowing are worse than the consequences of enabling. Here is one such case:

Hospital

In a hospital, a doctor has just plugged one person into the only available respirator. If the doctor either moves the one or unplugs him from the respirator, he will die. Five persons then arrive at the hospital and can be saved if and only if they are all plugged into the respirator at once. As it happens, the five will not survive being moved to the room with the respirator. But the respirator is movable.

In Hospital, the doctor has a choice between two options. The first option is to leave the one plugged into the respirator, in which case the one lives and the five die. The second option is to unplug the one from the respirator and plug the five into the respirator, in which case the one dies and the five live. (I ignore the clearly suboptimal option that involves unplugging the respirator from the one and then failing to attach it the five.)9 Under the circumstances, it seems clear that it is at least morally permissible (and perhaps even morally required) for the doctor to choose the second option over the first. The Equivalence Hypothesis provides a straightforward explanation of this result. On the first option, the doctor allows the five to die in that she fails to prevent whatever is ailing them from causing their deaths. On the second option, the doctor enables the one to die in that she removes an obstacle (the respirator) that is preventing whatever is ailing the one from causing his death. So the doctor is choosing between allowing five to die and enabling one to die. If enabling harm is morally equivalent to allowing harm, then the only morally relevant difference between the options is that choosing the first will lead to greater harm than choosing the second. Given that it is morally desirable to minimize

harm when all else is equal, it follows that it is morally permissible (and perhaps even morally required) for the doctor to choose the second option over the first. If the Equivalence Hypothesis is true, then in situations in which an agent is choosing between doing harm to one and enabling the same kind of harm to befall five, the DDA predicts that it is morally impermissible to choose the former option. For the DDA says that it is morally impermissible to do harm to one even if the alternative is to allow five to suffer the same kind of harm. But, according to the Equivalence Hypothesis, there is no morally relevant difference between allowing harm and enabling the same kind of harm. It follows directly from the conjunction of the DDA and the Equivalence Hypothesis that it is morally impermissible to do harm to one even if the alternative is to enable five to suffer the same kind of harm. Is this prediction borne out? Indeed it is. For consider the following case:

Motion Detector George is driving on a motion sensitive road. At the same time, not far from where George is driving, a large boulder is rolling down a hill towards five innocent people who are trapped in the boulders path. Luckily, there is a steel fence between the boulder and the five trapped people. If the boulder hits the fence, the five will be saved. However, George discovers that if he stops the car, the motion detector will set off an explosion that will destroy the fence before the boulder reaches it. In order to save the five, George resolves to keep driving. Unfortunately, George now discovers an additional innocent person trapped on the road ahead. If George keeps driving, he will crush and kill the one.

In Motion Detector, George is forced to choose between two options. The first option is to stop the car, in which case the one lives and the five die. The second option is to keep driving, in which case the one dies and the five live. Under the circumstances, it seems clear that it is morally impermissible for George to choose the second option over the first: he may not kill the one on his way to saving the five. In all morally relevant respects, then, Motion Detector is similar to Rescue, and this provides further confirmation for both the DDA and the Equivalence Hypothesis.

4. CHALLENGES TO THE EQUIVALENCE HYPOTHESIS The straightforward classification of harmful enabling as morally equivalent to harmful allowing, which is part and parcel of Foots original description and defense of the DDA, has recently come under attack. Various theorists, both consequentialists and nonconsequentialists, have argued that there are clear counterexamples to the Equivalence Hypothesis. Indeed, so they claim, there are many harmful enablings that are morally indistinguishable from harmful doings. For Foots non-consequentialist opponents, this is a sign that the DDA requires significant revision; for Foots consequentialist opponents, it is a sign that the DDA is unprincipled and should be abandoned.

McMahan Let us begin by looking at Jeff McMahans criticisms of the Equivalence Hypothesis. McMahans challenge begins with a case he calls Respirator:

Respirator A person is stricken with an ailment that would normally be fatal but is given mechanical life-support to sustain him until the condition can be cured. While the patient is on a respirator, his enemy surreptitiously enters the hospital and turns the machine off. The patient dies. (McMahan 1993, p. 254)

As McMahan sees it, if the distinction between enabling harm and doing harm marks an intuitively morally important difference, thenwe must conclude that it misclassifies this case for it is more natural to describe [Respirator] as a case of killing [that is, as a form of doing harm]; and we certainly evaluate it as such (McMahan 1993, p. 254). McMahan also offers the following potentially worrisome case:

Burning Building A person trapped atop a high building that is on fire leaps off. Seeing this, a firefighter quickly stations a self-standing net underneath and then dashes off to assist with other work. The imperiled persons enemy is, however, also present and, seeing his opportunity, swiftly removes the net so that the person hits the ground and dies. (McMahan 1993, p. 254)

In both Respirator and Burning Building the relevant victims enemy removes an obstacle that would otherwise prevent a causal sequence from leading to the victims death. So both cases are instances of enabling harm. But McMahans intuition in Burning Building, no less than in Respirator, is that enemys enabling victims death is a

killing that should be treated as morally equivalent to other killings (such as driving over the one on the second option in Rescue.) Although I share McMahans intuitions about both cases, I would argue that they provide no evidence for the falsity of the Equivalence Hypothesis. The problem in both cases is that the agent whose actions enable the victims death is described as the victims enemy. Persons who are described as enemies of those whose death they bring about intentionally are generally understood to be acting with malicious intent. So, when faced with both cases, we take it for granted that enemy acts with malice aforethought. Our intuitions being responsive to what we take for granted, they reflect our evaluation of malicious harmful enabling, rather than our evaluation of harmful enabling tout court. Our intuitions therefore count as evidence for the claim that malicious harmful enabling is morally equivalent to harmful doing, but they do not suggest that non-malicious harmful enabling is morally equivalent to harmful doing.10 The problem with both of McMahans initial counterexamples to the Equivalence Hypothesis is reminiscent of the problem with James Rachels well-known argument for the moral equivalence of doing and allowing.11 Rachels argument relies on a famous pair of cases. In the first case, Smith maliciously drowns his nephew to obtain an inheritance. In the second case, Jones maliciously fails to save his drowning nephew to obtain an inheritance. Rachels intuition is that Smiths action is as morally reprehensible as Joness inaction, and hence that there is no morally relevant difference between doing harm and allowing harm. But here too the most that could be said to follow from this pair of cases is that malicious harmful allowing is morally equivalent to malicious harmful doing.

In response to this reply, McMahan might argue that his purpose in describing Respirator and Burning Building is not to establish that some harmful enablings are morally equivalent to harmful doings. Rather, he might argue, his purpose is to do no more than contribute to the identification of the empirical criteria governing the use of the concepts of killing and letting die, and perhaps more broadly, the concepts of doing harm and allowing harm (McMahan 1993, p. 250). If this is correct, then the point of the two cases is to show that ordinary language classifies at least some enablings as killings, and more generally as harmful doings. If this is indeed the point of Respirator and Burning Building, then I wholeheartedly concur with McMahans conclusions about the application conditions of killing and doing harm in ordinary language. It seems perfectly appropriate to classify enemys action in Respirator as a killing, and as a case of doing harm; similarly for enemys action in Burning Building. But, from the moral point of view, which is, after all, what is driving McMahans investigation into the concepts of killing and letting die, these conclusions are neither here nor there. Foots main thesis is not a claim about the ordinary language application conditions of terms like killing and letting die; it is, rather, a claim about the moral significance of a technical distinction (between allowing and enabling on the one hand, and doing and sustaining on the other) that, as she perhaps incorrectly surmised, maps roughly onto the ordinary language distinction between allowing and doing. What matters morally is not whether (some) enablings are properly described as doings in the ordinary sense, but whether enabling is to be treated as morally equivalent to doing in the technical sense. And no investigation into the

10

application conditions of terms in ordinary language is capable of shedding light on this question. Let us then return to the moral question. Are there any non-malicious harmful enablings that we find to be morally indistinguishable from non-malicious harmful doings? McMahan, as it happens, considers a case of non-malicious harmful enabling, namely the following variant of Burning Building:

Burning Building 2 A person trapped atop a high building that is on fire leaps off. Seeing this, a firefighter quickly stations a self-standing net underneath. But he then immediately notices that two other persons have jumped from a window several yards away. He therefore repositions the net so that it catches the two. The first jumper then hits the ground and dies. (McMahan 1993, p. 262)

About this case, McMahan writes that it seems absurd to say that the firefighter kills the one; rather he merely allows him to die. Thats as may be. The moral question, however, is whether Burning Building 2 counts as evidence for, or perhaps even evidence against, the Equivalence Hypothesis. And, interestingly, the answer is that the case provides striking support for the hypothesis. In Burning Building 2, once he has placed the net underneath the first jumper, the firefighter faces a choice between two options. The first option is to leave the net where it is, in which case the two jumpers die and the first jumper lives. The second option is to reposition the net underneath the pair of jumpers, in which case the two jumpers live and the first jumper dies. On the first option,

11

the firefighter saves one but allows two to die. On the second option, he saves two but enables one to die. He will therefore save the greater number if he chooses option two. We may suppose that the firefighters choice is not guided by any malicious intentions; he is focused exclusively on helping. If non-malicious harmful enabling were morally equivalent to harmful doing, then, according to the DDA it would be morally impermissible for the firefighter to reposition the net to save the greater number. And yet, intuitively, the firefighter is at least morally permitted, and perhaps even morally required, to reposition the net. So Burning Building 2 strongly suggests that there is no morally relevant difference between harmful enabling and harmful allowing. In this respect it is strikingly similar to Hospital. The fact that McMahan interprets Burning Building 2 as a case of letting die, rather than a case of killing, suggests that he shares my moral intuitions about the case. But McMahan also claims that in all variants of the original Burning Building case in which the harmful enabling is performed by someone who did not place the net under the first jumper (or who did not occupy the same role as the person who placed the net under the first jumper), the relevant enabling counts as a killing. And this suggests that he might balk at the claim that it is morally permissible to enable harm to save the greater number in these variants. But this would be a mistake. For consider the following variant:

Burning Building 3 A person trapped atop a high building that is on fire leaps off. Seeing this, a firefighter quickly stations a self-standing net underneath. He then quickly leaves

12

the scene to assist those of his comrades who are battling the blaze. At the same time, Sally, a passer-by, then immediately notices that two other persons have jumped from a window several yards away. She therefore repositions the net so that it catches the two. The first jumper hits the grounds and dies.12

In this case, Sally faces the same options that the firefighter faces in Burning Building 2; she can allow one to be saved and allow two to die, or she can save the two and enable one to die. And it seems no less morally permissible for her to choose the second option in this case than it is for the firefighter to choose the second option in Burning Building 2. It may be that Sally would ordinarily be described as having killed the one if she chooses option two (though I myself do not share this linguistic intuition), but again what matters is not whether her enabling is properly described as a doing in the ordinary sense, but rather whether her enabling is to be treated as morally equivalent to a doing in the technical sense. And the intuitive answer to this question is clearly in the negative. McMahans theory of the factors that determine whether we will ordinarily count a fatal enabling as a killing or as a letting die is complex and fascinating. One of his claims is that it is relevant to the evaluation of a fatal enabling as a killing or as a letting die whether the obstacle that is removed is complete and self-sustaining, requiring no further contribution from [the one who originally provided it] to keep the threat at bay (McMahan 1993, p. 256). As evidence of this, McMahan offers us the following pair of cases:

The Pipe Sealer

13

An earthquake cracks a pipe at a factory, releasing poisonous chemicals into the water supply. Before a dangerous amount is released, a worker seals the pipe. But a year later he returns and removes the seal. As a result, numerous people die from drinking contaminated water. (McMahan 1993, p. 256)

The Dutch Boy A little Dutch boy, seeing that the dike is beginning to crack, valiantly sticks his finger in the crack to prevent the dike from breaking and flooding the town. He waits patiently but after many hours no one has come along who can help. Eventually succumbing to boredom and hunger, the boy withdraws his finger and leaves. Within minutes the dike bursts and a flood engulfs the town, killing many. (McMahan 1993, p. 257)

McMahan claims that our ordinary language intuitions bid us classify The Pipe Sealer as a case of killing but The Dutch Boy as a case of letting die. Again, thats as may be. The moral question, however, is whether the pipe sealers removal of the seal is morally equivalent to a harmful doing, and whether the Dutch boys removal of his finger is morally equivalent to a harmful allowing. And the answer here, it seems to me, is that both forms of enabling are, yet again, morally equivalent to harmful allowing, rather than to harmful doing. To see this, consider the following variants of both cases:

The Two Pipes

14

An earthquake cracks a pipe at a factory, releasing poisonous chemicals into a towns water supply. Before a dangerous amount is released, a worker seals the pipe. The worker realizes that if the seal is removed, one thousand people will die from drinking contaminated water. A little while later the worker returns to the factory to inspect the pipes. He notices a crack in a second pipe, which, if left unsealed, will release chemicals that will kill two thousand people in the next town over. Unfortunately, the only way to seal the crack in the second pipe is to remove the seal on the first pipe and transfer it to the second.

The Dutch Boy and the Trolley A little Dutch boy, seeing that the dike is beginning to break, valiantly sticks his finger in the crack to prevent the dike from breaking, flooding the town, and killing one thousand people. He then notices a runaway trolley with two thousand people in it, hurtling down a nearby hillside, close to going over the edge of a cliff and killing everyone in it. The boy realizes that if he pulls a lever a few yards away, the trolley will stop before it goes over the edge. But he knows that in order to reach the lever he will need to remove his finger from the crack, and that if he does so the dike will break.

In The Two Pipes the worker has two options. He can leave the seal on the first pipe, thereby saving one thousand people but allowing two thousand to die; alternatively, he can remove the seal and place it on the second pipe, thereby saving two thousand people but enabling one thousand to die. In The Dutch Boy and the Trolley the boy also has two

15

options. He can leave his finger in the dike, thereby saving one thousand people but allowing two thousand to die; alternatively, he can remove his finger from the dike and pull the lever to stop the runaway trolley, thereby saving two thousand people but enabling one thousand to die. If the harmful enabling in each of these cases were morally equivalent to harmful doing, then according to the DDA it would be morally impermissible for the worker to remove the seal from the first pipe and place it on the second, and it would also be morally impermissible for the Dutch boy to remove his finger from the dike and pull the lever. And yet, intuitively, these are actions that the worker and the boy are at least morally permitted, and perhaps even morally required, to perform. The upshot is that McMahans purported counterexamples establish neither that enabling harm per se is morally inequivalent to allowing harm per se, nor that enabling harm per se is morally equivalent to doing harm per se. Some of McMahans examples could at best be used to infer that malicious harmful enabling is morally equivalent to harmful doing. The rest of his examples show at best that particular cases of enabling harm can, under certain circumstances, fall under the ordinary language concept of doing harm. The moral is that McMahan has given us no good reason to abandon Foots original claim that there is no morally significant difference between harmful enabling and harmful allowing.

2. Vihvelin and Tomkow McMahan is not the only non-consequentialist opponent of the Equivalence Hypothesis. Recently, Vihvelin and Tomkow (henceforth, V&T) have argued that some cases of

16

enabling harm are morally equivalent to doing harm. As their paradigms of doing harm and allowing harm, V&T use Bennetts examples, Push and Stayback:

Push A cart stands at the top of a hill. Agent pushes it. The cart rolls down the hill and fatally injures a child.

Stayback The cart is already rolling; Agent could but does not interpose a rock which would stop it. The cart rolls down the hill and fatally injures a child.13

V&T then consider the following examples of enabling harm, the first of which is Bennetts:

Kick The cart is rolling towards a point where there is a rock that would bring it to a halt. Agent kicks away the rock; the cart rolls down the hill and fatally injures a child.

Dislodge A cart stands at the top of a hill, its wheel chocked by a rock. Agent kicks the rock away. The cart rolls down the hill and fatally injures a child.14

17

V&T describe Kick as a case of preventer prevention (for in kicking away the rock, Agent prevents the rock from preventing the childs death), and describe Dislodge as a case of causation by disconnection.15 Regarding these two cases of enabling harm, V&T argue as follows:

[W]hatever temptation there may be to use the language of allowing in [Kick and Dislodge] there is no corresponding temptation to treat these agents morally on a par with the agent of Stayback. In real life cases of disconnection and preventer preventionthe hangman springing the gallows trap, the vandal who removes the guard rail at the tourist lookout or the greedy pharmacist who waters the vaccinewe count the agents as killers on a par with Push and not bystanders like Stayback. (Vihvelin and Tomkow 2005, p. 194)

V&T conclude that at least some types of fatally harmful enabling, though perhaps not all, are morally equivalent to killing, and, more generally, that at least some types of harmful enabling are morally equivalent to harmful doing. But it should now be easy to see why V&Ts argument does not establish this conclusion. V&T claim that Agents kicks in Kick and Dislodge are actions that we count as killings. But, again, whether we count them as killings in ordinary language is irrelevant to the question of whether they are morally equivalent to harmful doings in the technical sense. What, then, supports V&Ts claim that Agents kicks in Kick and Dislodge are morally inequivalent to harmful allowing? The answer is that, on their

18

view, these kicks are analogous to other cases of disconnection and preventer prevention, each of which is morally equivalent to harmful doing. However, when we look at these cases more carefully, we can see that they do not establish what V&T think they do. V&T mention three cases. Consider the last two, the vandal who removes the guard rail at the tourist lookout and the greedy pharmacist who waters the vaccine. (I discuss the case of the hangman springing the gallows trap in the following section.) Now, ordinarily, vandals are understood to be persons who have little or no regard for public or private property. Moreover, any person who removes a guard rail at a lookout knows, or should know, that the action could prove fatal to tourists who stop by. So removing the guard rail is grossly negligent at best, and malicious at worst. Further, a greedy pharmacist who waters the vaccine (presumably without informing the people who might be inoculated with it) is willing to trade the lives of innocent persons for his own benefit, and he too is guilty of gross malfeasance at best, malicious wrongdoing at worst. The fact that the intentions of the relevant agents in these two cases are not directed at the good serves as a distortion in the eliciting of intuitions about the moral status of harmful enabling per se. In this respect, these cases are similar to McMahans Respirator and Burning Building, cases in which the relevant agent is described as the enemy of the relevant victim. We can no more legitimately infer from the case of the vandal and the case of the greedy pharmacist that certain sorts of harmful enabling are morally equivalent to harmful doing than we can infer the same conclusion from Respirator and Burning Building. The most that could possibly be inferred from the vandal case and the greedy pharmacist case is that negligent, malfeasant, or malicious

19

harmful enabling is morally equivalent to negligent, malfeasant, or malicious harmful doing. And this is not sufficient to establish what V&T want to show. Another case that is sometimes used to support the claim that some types of harmful enabling are morally equivalent to harmful doing is the following scenario:

Drive Away Suppose As car is parked between a rolling rock and a helpless child. A sees the child in the rocks path, but drives off to avoid an ugly dent.16

Boorse and Sorensen (henceforth, B&S) claim that A can be convicted of murder or manslaughter in any American court, and conclude that there is no morally significant difference between this kind of enabling and a doing that leads to the same result (Boorse and Sorensen 1988, p. 127). V&T claim that [t]he driver is as much as killer as the agent in Push, which is why the juries would convict (Vihvelin and Tomkow 2005, p. 203). But, again, from the moral point of view, whether we are inclined to describe the case as a killing is neither here nor there. It is, of course, morally reprehensible for A to drive away to avoid a dent. But it is also morally reprehensible for A (in a slightly different scenario) to fail to interpose his car between the rolling rock and the child if his reason for failing to do so is to avoid an ugly dent. And in this sort of case A allows the child to die. So the fact that As action is morally reprehensible in Drive Away does not show that harmful enabling is morally inequivalent to harmful allowing. What matters is the question of moral equivalence, and here again it is easier to answer this question by altering the case so that it involves a choice between enabling

20

one to die and saving five or allowing five to die and saving one. What we need is a case much like Burning Building 2. Lets call it Drive Away 2:

Drive Away 2 As car is parked between a rolling rock and one helpless child. A now sees a second rock rolling towards a group of five other helpless children. If A leaves his car where it is, the one helpless child will be saved but the five helpless children will die. But A can drive a short distance and stop, thereby interposing his car between the second rolling rock and the five helpless children. If A does this, the five helpless children will be saved but the one helpless child will die.

In Drive Away 2, it seems morally permissible for A to drive the short distance and stop his car to save the five. Yet in doing so he enables one helpless child to die. So if harmful enabling were morally equivalent to harmful doing, then the DDA would speak against his moving the car. Drive Away 2 therefore suggests that harmful enabling is morally equivalent to harmful allowing, rather than to harmful doing.

3. Gallows and Guns Thus far, we have considered and rejected a number of putative counterexamples to the Equivalence Hypothesis. But there are two more particularly recalcitrant types of counterexample that some cite as evidence for the claim that at least some types of harmful enabling are morally equivalent to harmful doing, and hence morally

21

inequivalent to harmful allowing. I now want to explain why these putative counterexamples do not, in the end, doom the Equivalence Hypothesis. Let us return to V&Ts case of the hangman springing the gallows trap. We may suppose, for arguments sake, that the hangman has excellent moral reasons for springing the trap. Perhaps the person hanged was himself an unrepentant mass murderer. Isnt this scenario a case in which harmful enabling is morally equivalent to harmful doing? After all, in springing the trap, the hangman simply removes an obstacle with a fatal upshot (namely, the victims fall into the void, ending with the tightening of the noose around his neck and consequent death by asphyxiation). And isnt the hangmans action morally equivalent to asphyxiating the unrepentant murderer directly? Perhaps so. But again, the assumption that the hanged person is an unrepentant murderer, and hence deserves to be hanged, counts as a distorting element in the eliciting of moral intuitions about case. When someone deserves death, it may not matter morally whether his death is enabled or brought about directly. But it does not follow that there is no morally relevant difference between harmful enabling per se and harmful doing per se. To see whether there is such a morally relevant difference, it helps to change the case to one in which the person who is in danger of being hanged is innocent of wrongdoing and finds himself with a noose around his neck through no fault of his own. Then, to test the Equivalence Hypothesis, we can alter the circumstances further by adding a choice between springing the trap door and allowing a greater number of people to be hanged. Here is such a case:

The Two Trap Doors

22

One innocent person finds himself standing on a trap door with a noose around his neck. At the same time, five innocent people with nooses tied to long ropes around their necks are falling towards a second heavily padded, but open trap door. The two trap doors are mechanically connected so that if the first trap door is sprung, the second automatically closes. Roger, a bystander, happens upon the lever that will spring the first trap door. Roger knows that if he pulls the lever, the first trap door will open and one innocent person will be hanged, while at the same time the second heavily padded trap door will close, and five innocent people will be saved. On the other hand, Roger knows that if he doesnt pull the lever, then the second trap door will remain open and five innocent people will be hanged, while at the same time the first trap door will remain closed and one innocent person will be saved.

Roger has a choice between bringing about the death of one while saving five or allowing five to die while allowing one to be saved. And it does indeed appear that Roger is not permitted to pull the lever that opens the first trap door, thereby sending the one innocent person to his death, even if by doing so he saves five innocent lives. Does this sort of case not show that at least one kind of harmful enabling is morally equivalent to harmful doing rather than to harmful allowing? Now I said above (in footnote 6) that I would not attempt to provide a complete account of the metaphysics of causal sequences. But even without such an account, it is possible to see that The Two Trap Doors does not actually involve a choice between allowing harm and enabling harm. The crucial point is that in The Two Trap Doors there

23

is no pre-existing causal sequence that will harm the one if the relevant lever is pulled. We are imagining that the one is immobilized on a trap door with a noose around his neck, a noose that will tighten and asphyxiate the one if the trap door is opened. The opening of the trap door does not make it possible for a causal sequence that is already in train to continue, for there is no such causal sequence. In particular, the one is not in the process of falling. The situation is, rather, completely stable.17 Thus, as seems clear, the opening of the trap door constitutes the initiation of a causal sequence that leads to the death of the one, namely his falling through the air and consequent tightening of the rope around his neck. This is why we judge it morally impermissible to spring the first trap door, for the initiation of a fatal sequence counts as a harmful doing, and the DDA does not permit us to do harm to one even if this enables us to avoid allowing (the same kind of) harm to five. To see that The Two Trap Doors actually involves the initiation, rather than the enabling, of a fatal causal sequence, it may help to contrast the case with a similar case in which both trap doors are flimsy and in the process of cracking. Consider:

The Two Flimsy Trap Doors One innocent person finds himself standing on a flimsy trap door with a noose around his neck. The trap door is in the process of cracking under the persons weight. At the same time, not far away, five innocent people with nooses around their necks find themselves standing on a second flimsy trap door that is in the process of cracking. Luckily for the one, pointing at the first trap door is a special laser, the energy from which is preventing the door from breaking in half and

24

giving way. The laser can point in one of only two directions: towards the first trap door or towards the second. Rita, a bystander, happens upon the laser. Rita knows that if she points the laser away from the first trap door towards the second, the first trap door will break and one innocent person will be hanged, while at the same time five innocent people will be saved (because the energy from the laser will prevent the second door from giving way). On the other hand, Rita knows that if she does nothing, then the second trap door will give way and five innocent people will be hanged, while at the same time the first trap door will not give way and one innocent person will be saved.

Rita has a choice between leaving the laser alone while it is pointing at the first trap door or and moving it so that it is pointing at the second trap door. If she does nothing, the five will die but the one will live; if she moves the laser, the five will live but the one will die. Here it seems morally permissible for Rita to move the laser. But, as seems plain, there is a pre-existing causal sequence (the first trap door in the process of cracking under the ones weight), the potentially fatal upshot of which is being staved off by the laser. So when Rita moves the laser so that it points to the second trap door instead of the first, she removes an obstacle to a potentially fatal pre-existing sequence, and hence merely enables (rather than does) harm. The difference in our moral intuitions about the two cases (that it is morally impermissible for Roger to pull the lever in The Two Trap Doors, while it is morally permissible for Rita to move the laser in The Two Flimsy Trap Doors) is easily explained by the fact that Roger would do harm in pulling the lever, while Rita would only enable harm in moving the laser.

25

One might object to my treatment of these cases that The Two Trap Doors is, in all causally relevant respects, isomorphic to both The Pipe Sealer and The Dutch Boy.18 In The Two Trap Doors, the one has a rope around his neck and gravity is pulling him towards the floor beneath the trap door. One might be tempted to say that all that stands in the way of the realization of this sequence is a trap door that is removed by the agent. Moreover, the case seems relevantly like the case of The Dutch Boy, in which all that stands in the way of the flood is the boys finger in the dike, and also relevantly like the case of The Pipe Sealer, in which all that stands in the way of death by water contamination is a sealant. But, in fact, The Two Trap Doors is not relevantly similar to The Dutch Boy and The Pipe Sealer. Recall that in The Dutch Boy, as McMahan describes it, the dike is beginning to crack and the boy sticks his finger in the crack to prevent the dike from breaking. Similarly, in The Pipe Sealer, an earthquake has already produced a crack in a pipe, thereby releasing poisonous chemicals into the water supply, and the pipe sealer seals the pipe before a dangerous amount is released. There is therefore no difficulty in identifying a pre-existing causal sequence in The Dutch Boy. The pressure of the water behind the dike is producing cracks in the dike, cracks that will result in the dikes giving way unless something is placed in the hole produced by the cracks to prevent further cracking. Similarly, there is no difficulty in identifying a preexisting causal sequence in The Pipe Sealer, for it is clear that toxic chemicals are already being released into the water supply through a hole produced by a crack in a pipe. The Dutch Boy and The Pipe Sealer are therefore relevantly similar to The Two Flimsy Trap Doors, and relevantly dissimilar from The Two Trap Doors.19

26

So much for the case of springing the gallows trap. But there is a second type of putative counterexample to the Equivalence Hypothesis. The example I have in mind is one in which one person shoots another. Frances Howard-Snyder considers the following scenario:

Sassan shot Victor. He pulled the trigger. The gun fired. A bullet flew out of the barrel and entered Victors body. Victor died from the bullet wound. A clearer case of killing is impossible to find. (Howard-Snyder 2002, section 6)

According to Howard-Snyder, it is obvious that Sassan kills Victor by shooting him. And indeed it does seem obvious. However, as Schaffer points out, pulling the trigger of a (certain sort of) gun involves disconnecting the sear, allowing the spring to uncoil (propelling the striker onto the powder, compression of which produces the explosion which propels the bullet) (Schaffer 2000, p. 287). Unless the trigger is pulled, the sear prevents the explosion that leads to the firing of the bullet. So pulling a guns trigger appears to be a kind of enabling, for it involves the removal of an obstacle that lies in the way of an explosion that propels the fatal bullet. So when Sassan shoots Victor, his killing of Victor appears to be an enabling. Moreover, it seems obvious, not only that Sassan kills Victor, but also that Sassans killing of Victor is morally equivalent to doing harm. For there seems to be no morally relevant difference between Sassans shooting Victor on the one hand and Sassans pulverizing Victor on the other. However, the Sassan case does not falsify the Equivalence Hypothesis. First, as we saw earlier, the fact that Sassan kills Victor does not entail that harmful enabling is

27

morally equivalent to harmful doing. Because doing harm is a technical concept, it does not follow from the fact that something is correctly described in ordinary language as a killing that it counts as a doing. Nor does the fact that many or most killings are doings entail that all killings are doings. So there is no inconsistency in saying both that Sassan kills Victor and that he does not do harm (in the technical sense) in killing Victor. Second, it is easy to explain why there appears to be no morally relevant difference between Sassans shooting Victor and Sassans pulverizing Victor. In both cases, it is obvious that Sassan does something wrong, indeed something morally reprehensible. It may therefore seem to follow that the type fatal shooting is morally indistinguishable from the type fatal pulverizing. But it does not follow. This is the familiar point brought out by the failure of Rachels Smith-Jones pair of cases. Just as the fact that Joness allowing his cousin to drown is just as morally reprehensible as Smiths drowning his cousin does not entail that allowing harm is morally equivalent to doing harm, so the fact that Sassans shooting Victor is just as morally reprehensible as Sassans pulverizing Victor does not entail that enabling harm is morally equivalent to doing harm. The confounding parameter in both pairs of cases is the same, namely the malicious intent of the relevant agent of harm. To correctly gauge the moral status of fatal shootings, it helps to consider another scenario in which the agent has a choice between shooting one while saving five and allowing five to die while one is saved. Here is one such case:

The Gun and the Bomb

28

A bomb is set to go off in five minutes. Five innocent people are strapped to the bomb, and all five will die if the bomb goes off. Several yards away, there is a gun encased in concrete, pointed at the heart of one innocent person, also encased in concrete some distance away. If the gun is fired, the one innocent person will die. Reina, a savvy bystander who has just arrived on the scene, notices that the guns trigger is linked to the bombs detonator by a motion-detector: if the trigger remains motionless, then the detonator will fire and the bomb will explode; but if the trigger is pulled, thereby allowing the propulsion of the striker onto the powder and the consequent explosion that propels the bullet towards the one, then the detonator will be deactivated and the bomb rendered harmless.

Reina has a choice. On the one hand, she can leave the trigger alone, in which case one lives but five die. On the other hand, she can pull the trigger, in which case one dies but five live. Is Reina morally permitted to pull the trigger? My own intuition is that Reina is not permitted to pull the trigger. But isnt pulling the trigger the removal of an obstacle to a sequence that leads to the explosion that propels the bullet into the body of the one? And doesnt this mean that pulling the trigger is a kind of harmful enabling? If so, then according to the DDA and the Equivalence Hypothesis, it should be morally permissible to pull the trigger. But here again the question arises whether pulling the trigger makes it possible for a pre-existing causal sequence to lead to harm. And the answer, as I see it, is that it does not. The reason for this is simple: although pulling the trigger does indeed release the potential energy bottled up in the coiled spring, it does not thereby enable a pre-existing

29

causal sequence to lead to harm. Rather, the trigger pulling initiates the causal sequence that leads to the ones death. The reason for this is similar to the reason for thinking that there is no pre-existing sequence in The Two Trap Doors. In the latter case, the trap door is locked, while in The Gun and the Bomb the trigger is locked. The situation in both cases is completely stable.20 The moral of the gallows trap and the trigger is that not all cases of causation by disconnection are morally equivalent. There are cases in which an obstacle stands in the way of a pre-existing causal sequence that will lead to harm is the obstacle is removed; and there are cases in which an obstacle prevents the initiation of a causal sequence that will lead to harm if the obstacle is removed. The former cases are cases of enabling harm that are morally equivalent to cases of allowing harm; the latter cases are cases of initiating harm.

5. CONCLUSION In recent years, it has become increasingly common to assume, on the basis of simple (non-forced-choice) cases of disconnection and preventer prevention, that some harmful enablings are morally equivalent to harmful doings. Indeed, several authors have built (or have suggested) their own rather complex versions of the DDA on this assumption.21 As I have argued, none of these cases succeeds in disproving Foots original claim that the moral status of enabling harm is equivalent to the moral status of allowing harm. We are left to conclude that there is no morally relevant difference between harmful enabling and harmful allowing per se, and hence that the more complex versions of the DDA that are based on the rejection of this conclusion are false.22

30

Department of Philosophy University of California, San Diego

31

REFERENCES

Bennett, Jonathan. The Act Itself. (Oxford: Clarendon Press, 1995).

Boorse, Christopher and Roy A. Sorensen. (1988). Ducking Harm, Journal of Philosophy 85, pp. 115-134.

Dowe, Phil. (1996/2007). Causal Processes, In Stanford Encyclopedia of Philosophy, E. N. Zalta (ed), URL= <http://plato.stanford.edu/entries/causationprocess/#AroTraThe>

Foot, Philippa. (1967). The Problem of Abortion and the Doctrine of Double Effect, Oxford Review 5, pp. 5-15. Reprinted in Virtues and Vices and Other Essays in Moral Philosophy (Oxford: Blackwell, 1978), pp. 19-32, and in Steinbock and Norcross (1994), pp. 266-279.

Foot, Philippa. (1984). Killing and Letting Die. In Abortion: Moral and Legal Perspectives, J. L. Garfield and P. Hennessey (eds), (Amherst: University of Massachusetts Press), pp. 177-183. Reprinted in Steinbock and Norcross (1994), pp. 280-289.

Foot, Philippa. (1985). Morality, Action and Outcome. In Morality and Objectivity: A

32

Tribute to J. L. Mackie, T. Honderich (ed), (London: Routledge & Kegan Paul), pp. 23-38.

Hanser, Matthew. (1999). Killing, Letting Die and Preventing People from Being Saved, Utilitas 11, pp. 277-295

Howard-Snyder, Frances. (2002). Doing vs. Allowing Harm, Stanford Encyclopedia of Philosophy, E. N. Zalta (ed), URL=<http://plato.stanford.edu/entries/doingallowing/>

Kagan, Shelly. The Limits of Morality. (Oxford: Clarendon Press, 1989).

McMahan, Jeff. (1993). Killing, Letting Die, and Withdrawing Aid, Ethics 103, pp. 250-279.

Rachels, James. (1975). Active and Passive Euthanasia, New England Journal of Medicine 292, pp. 78-80. Reprinted in Steinbock and Norcross (1994), pp. 112119.

Schaffer, Jonathan. (2000). Causation by Disconnection. Philosophy of Science 67, pp. 285-300.

Steinbock, Bonnie and Alastair Norcross (eds.). Killing and Letting Die, 2nd

33

Edition. (New York: Fordham University Press, 1994).

Unger, Peter. Living High and Letting Die: Our Illusion of Innocence. (Oxford: Oxford University Press, 1996).

Vihvelin, Kadri and Terrance Tomkow. (2005). The Dif, Journal of Philosophy 102, pp. 183-205.

Woollard, Fiona. (2008). Doing and Allowing, Threats and Sequences, Pacific Philosophical Quarterly 89, pp. 261-277.

34

ENDNOTES

See Foot (1967) and Hanser (1999).

See Boorse and Sorensen (1988), McMahan (1993), Vihvelin and Tomkow (2005), and

Woollard (2008).

See Rachels (1975), Kagan (1989), Bennett (1995), Unger (1996), and Howard-Snyder

(2002).

There are other puzzles for non-consequentialism, including puzzles concerning the

proper rationale for treating the distinction between doing harm and allowing harm as morally significant, and puzzles concerning the causal metaphysics behind the distinction (for example, whether there is a metaphysical basis for distinguishing between enabling harm and sustaining harm). These puzzles require special treatment, and I do not propose to solve them here.

See Foot (1967; 1984; 1985).

To provide paradigm cases of initiating, sustaining, and allowing is not yet to provide

an analysis of these phenomena. An analysis will require a complete theory of causation and of causal sequences. Providing such an analysis is beyond the scope of this paper.

35

But it should be noted that Foots account works best with causal process accounts of causation and causal sequences, such as the mark transmission theory defended Wesley Salmon or the conserved quantity theory defended by Phil Dowe. For details and an extensive bibliography, see Dowe (1996/2007).

The reason for the qualification is this. In situations in which the potential victims

embrace the relevant harm (e.g., in cases of voluntary euthanasia), it may be permissible to do harm rather than allow harm in situations of forced choice. Henceforth, I will drop the explicit qualification to avoid unnecessary complication. But it is implicit in all the cases to be discussed.

This case is inspired by the famous Rescue II case described in Foot (1984).

I will also ignore clearly suboptimal options in all the cases I go on to discuss.

10

A reviewer for this journal objects to this line of reasoning by means of an example

that involves a slight emendation to the classic Trolley case. Suppose, says the reviewer, that Brown is faced with a choice between turning the trolley onto one or letting the trolley kill five. Now introduce the wrinkle that Brown hates the one and wants him dead, and turns the trolley out of malice. The reviewers intuition (and the intuition of the reviewers students) is that this is a case of a right action done for the wrong reason. If this is the case, then, says the reviewer, it seems untrue that malicious enabling is as wrong as killing.

36

Now I myself do not happen to share the reviewers intuitions about the case. But leaving that to one side, my main response to this objection is that, contrary to the reviewers suggestion, I am not in fact arguing for, nor am I asserting, that malicious harmful enabling is equivalent to harmful doing. My point is merely that cases such as Respirator and Burning Building, as McMahan describes them, do not establish that harmful enabling tout court is morally equivalent to harmful doing. I claim that if these cases are evidence for anything, they are evidence for the more restricted proposition that malicious harmful enabling is morally equivalent to harmful doing. It may be, as the reviewer suggests, that this more restricted proposition is false. But this is compatible with the point I am making, which is that the evidence against the Equivalence Hypothesis represented by McMahans two cases is poor.

11

See Rachels (1975).

12

This case resembles the case of Two Cars and a Rock described by Hanser (1999, p.

285).

13

See Bennett (1995, p. 67), and Vihvelin and Tomkow (2005, p. 192).

14

See Bennett (1995, p. 67), and Vihvelin and Tomkow (2005, p. 193).

15

See Schaffer (2000).

37

16

See Boorse and Sorensen (1988, p. 127), and Vihvelin and Tomkow (2005, pp. 202-

203).

17

As causal process theorists would analyze the situation, the situation in The Two Trap

Doors does not involve the propagation of a mark (before the trap door is sprung), nor does it involve the transfer of any conserved quantity (see note 6).

18

I thank a reviewer for this journal for raising this useful objection.

19

Again, the mark-propagation theory and the conserved quantity transfer theory of

causal processes entail that there are pre-existing causal sequences at play in The Dutch Boy, The Pipe Sealer, and The Two Flimsy Trap Doors, but no such sequence in play in The Two Trap Doors (for details, see the references discussed in note 6).

20

Contrast The Gun and the Bomb with a case in which a bullet happens to be on its way

to killing one innocent person, but will be stopped by a metal plate unless Roberta pulls a lever that both pushes the plate out of the bullets way and places it in position to stop five other bullets that will kill five other innocents. If Roberta pulls the lever, five live but one dies. If Roberta does not pull the lever, one lives and five die. It seems clear that Roberta is morally permitted to pull the lever. But the pulling of the lever does not represent the initiation of the causal sequence that leads to the death of the one, for the lever-pulling simply makes it possible for a pre-existing sequence (the discharging of a firearm leading to the propulsion of a bullet towards the one) that would otherwise be

38

stopped to run its fatal course. By contrast, in The Gun and the Bomb, there is no causal sequence already in train that will lead to the death of the one unless Reina removes some obstacle that will, unless she removes it, stop the sequence before it results in anyones death. Rather, Reina starts the causal process that leads to the death of the one when she pulls the trigger. Again, this judgment is in line with causal process theories of causation and causal sequences (see note 6).

21

See, for example, McMahan (1993), Vihvelin and Tomkow (2005), and Woollard

(2008).

22

I would like to thank an anonymous reviewer and the editors of this journal for their

helpful and constructive comments. I would also like to thank Dana K. Nelkin for her invaluable advice and suggestions.

39